Majorization-minimization generalized Krylov subspace methods for \({\ell _p}\)-\({\ell _q}\) optimization applied to image restoration (Q2359752)

This paper considers the efficient computation of approximate solutions of \(\ell _p\)-\(\ell _q\) optimization problems that are applied to image restoration. The strategy used to generate adaptive or fixed quadratic majorants is described. A majorize-minimize generalized Krylov subspace algorithm is proposed by combining the majorize-minimize framework with generalized Krylov subspace methods. The convergence of the algorithm is established. Computed examples show that the proposed algorithm is more efficient than the current iteratively reweighted norm algorithm.